DIY individual stock fund

Discussion of the Stock portion of the Permanent Portfolio

Moderator: Global Moderator

User avatar
Tyler
Executive Member
Executive Member
Posts: 1943
Joined: Sat Nov 12, 2011 3:23 pm
Contact:

Re: DIY individual stock fund

Post by Tyler » Mon Sep 23, 2019 8:23 pm

Pet Hog wrote:
Mon Sep 23, 2019 7:47 pm
Thanks for all the comments. It's just something I'm mulling. Years ago I read on the Motley Fool that when your portfolio becomes enormous (I wish it were so) you should become your own index. And I think it was Machine Ghost on these forums who first got me interested in equal-weighting a portfolio. And then there's Tyler's Golden Butterfly, which adds small-cap value to the PP -- that's also a tilt toward equal-weighting, if you think about it.
Yep! Read some of my old posts on the GB and you may see that I also like small cap blend in the portfolio basically for the equal weighting potential you point out. Since total market funds also contain mid and small caps, and small cap funds also contain more mid caps than you'd think, mix 50% VTI and 50% VB and you basically evenly fill out a Morningstar style box. It's not quite as precisely equal weight as RSP, but it's also 1/5th the cost. 8)
Mechanical engineer, history buff, treasure manager... totally not Ben Gates
User avatar
KevinW
Executive Member
Executive Member
Posts: 945
Joined: Sun May 02, 2010 11:01 pm

Re: DIY individual stock fund

Post by KevinW » Mon Sep 23, 2019 11:56 pm

It is common for a company's size ranking to bump up or down one rank from year to year. So the method of picking every k'th company will experience high turnover and suffer from tax inefficiency and transaction overhead.

Probably the way to go is to just pick the top 50 (or whatever) largest companies and leave it at that. That's dead simple, will be highly correlated to the entire market, and will have low turnover.

Or, there is a classic technique in computer algorithms of using just a little bit of randomization. For example take the top 45 stocks by size, and then another 5 stocks chosen entirely at random. This manages to give you a reasonable chance of exposure to opportunities outside of the mega-caps while keeping overhead under control.

It's fun to think about this stuff but as a practical matter it's tough to beat funds like VTI or FZROX. These are really superb consumer products from the standpoint of expenses, tax efficiency, liquidity, return on time investment, and potential for human error.
boglerdude
Executive Member
Executive Member
Posts: 607
Joined: Wed Aug 10, 2016 1:40 am
Contact:

Re: DIY individual stock fund

Post by boglerdude » Tue Sep 24, 2019 1:38 am

> How much can the FAANG stocks really grow from here when they already have trillion-dollar market caps?

This is Arnott's shtick, check out the RAFI funds

https://ritholtz.com/2018/07/mib-gettin ... ob-arnott/
Pet Hog
Executive Member
Executive Member
Posts: 218
Joined: Tue May 28, 2013 4:08 pm

Re: DIY individual stock fund

Post by Pet Hog » Tue Sep 24, 2019 2:13 am

Tyler wrote:
Mon Sep 23, 2019 8:23 pm
Yep! Read some of my old posts on the GB and you may see that I also like small cap blend in the portfolio basically for the equal weighting potential you point out. Since total market funds also contain mid and small caps, and small cap funds also contain more mid caps than you'd think, mix 50% VTI and 50% VB and you basically evenly fill out a Morningstar style box. It's not quite as precisely equal weight as RSP, but it's also 1/5th the cost. 8)
I shall mull over a VTI/VB blend. Thanks, Tyler.
KevinW wrote:
Mon Sep 23, 2019 11:56 pm
It is common for a company's size ranking to bump up or down one rank from year to year. So the method of picking every k'th company will experience high turnover and suffer from tax inefficiency and transaction overhead.
Thanks for the wise comments, Kevin. I didn't make myself clear. My intention is to buy a set of stocks, maybe 40 or 50, and hold them indefinitely, as a proxy for the total market. Kind of like the DJIA or the Nifty Fifty. I wouldn't keep replacing the k-th company annually or every time I rebalanced. Just buy them all once at the initial purchase. Any dividends, or new funds, or rebalance funds out of gold/treasuries might go to a new and different stock, so the proxy index would grow over time to incorporate a greater number of companies. I really want to minimize turnover. In fact, I think I would let the losers lose and only redistribute gains from really big winners, like stocks that triple or more, and use those gains to buy holdings in entirely new companies. It's not important that everything be equally weighted to the penny. I just don't want a trillion-dollar company and a billion-dollar company being represented at a 1000:1 dollar ratio in my portfolio, as they are in VTI and such. A 2:1 or 3:1 ratio would be tolerable. Close to equal-weighted, but no stress.

Yes, it is fun to think about these things. I probably won't follow through with this scheme -- I'm happy with how VTI and ITOT have been treating me with little effort -- but I was hoping I might get some encouragement from my PP cohort. So far, none!
User avatar
dualstow
Executive Member
Executive Member
Posts: 10827
Joined: Wed Oct 27, 2010 10:18 am
Contact:

Re: DIY individual stock fund

Post by dualstow » Tue Sep 24, 2019 9:25 am

Pet Hog wrote:
Mon Sep 23, 2019 3:03 pm
Here, for example, is the list of every 16th company in the SP500:

1 Home Depot (market cap: $245 billion)
..
31 Discovery (market cap: $13 billion)

I think I could go for that. Seems diverse enough. Thoughts?
In a set of only 31 stocks, I count 3 banks. Home Depot *and* Lowe’s? What if you cheated a little by tweaking it or rerunning it until you have a better representation of different sectors?

Btw, I’m also a big fan of small cap blend. I’ve held VSMAX for a very long time (because I like Vanguard. I don’t think it’s a popular choice, even at Bogleheads). I hold a lot of large companies as individual stocks, but there are too many little ones, so it makes more sense to cast a wide net with an index fund.
The new Freakonomics: Why Are Cities (Still) So Expensive? (Ep. 435)
Pet Hog
Executive Member
Executive Member
Posts: 218
Joined: Tue May 28, 2013 4:08 pm

Re: DIY individual stock fund

Post by Pet Hog » Tue Sep 24, 2019 1:36 pm

dualstow wrote:
Tue Sep 24, 2019 9:25 am
In a set of only 31 stocks, I count 3 banks. Home Depot *and* Lowe’s? What if you cheated a little by tweaking it or rerunning it until you have a better representation of different sectors?
The Dow Jones Industrial Average comprises 30 companies. Including American Express, Goldman Sachs, and JP Morgan Chase. Three banks! Pfizer, Merck, and Johnson and Johnson? Three pharmaceutical companies! Chevron *and* Exxon? I could go on...

That list of 31 companies was selected purely randomly. I was just thinking out loud. Obviously, there would be tweaks if I ever used any of those selection strategies, but I wanted to see how mechanically and unemotionally I could assemble a reasonable total market proxy.

I should just model some of those portfolios and see how well they do with respect to something like VTI. The proof of the pudding. Let's see...
User avatar
dualstow
Executive Member
Executive Member
Posts: 10827
Joined: Wed Oct 27, 2010 10:18 am
Contact:

Re: DIY individual stock fund

Post by dualstow » Tue Sep 24, 2019 2:08 pm

Pet Hog wrote:
Tue Sep 24, 2019 1:36 pm
dualstow wrote:
Tue Sep 24, 2019 9:25 am
In a set of only 31 stocks, I count 3 banks. Home Depot *and* Lowe’s? What if you cheated a little by tweaking it or rerunning it until you have a better representation of different sectors?
The Dow Jones Industrial Average comprises 30 companies. Including American Express, Goldman Sachs, and JP Morgan Chase. Three banks! Pfizer, Merck, and Johnson and Johnson? Three pharmaceutical companies! Chevron *and* Exxon? I could go on...
True, but the Dow is a piss-poor representation of the market and not something that you want to emulate!
The only good thing about buying the Dow and nothing else would be that you could look up from your meal at the screen of a sports bar and see how your investments are doing. No spreadsheet needed.
The new Freakonomics: Why Are Cities (Still) So Expensive? (Ep. 435)
User avatar
ochotona
Executive Member
Executive Member
Posts: 3177
Joined: Fri Jan 30, 2015 5:54 am

Re: DIY individual stock fund

Post by ochotona » Tue Sep 24, 2019 3:12 pm

Only 8.6% of the Wilshire 5000 is small cap, 91.4% is large cap. Is that right?
Pet Hog
Executive Member
Executive Member
Posts: 218
Joined: Tue May 28, 2013 4:08 pm

Re: DIY individual stock fund

Post by Pet Hog » Tue Sep 24, 2019 3:52 pm

dualstow wrote:
Tue Sep 24, 2019 2:08 pm
True, but the Dow is a piss-poor representation of the market and not something that you want to emulate!
The only good thing about buying the Dow and nothing else would be that you could look up from your meal at the screen of a sports bar and see how your investments are doing. No spreadsheet needed.
Ha!
ochotona wrote:
Tue Sep 24, 2019 3:12 pm
Only 8.6% of the Wilshire 5000 is small cap, 91.4% is large cap. Is that right?
Probably, as measured by market cap. I think most of the entries are small/micro, as measured by number of companies.

I've added charts from portfoliovisualizer.com for all of the portfolios I listed yesterday. Buy and hold with dividends reinvested. Pretty good returns over the last 3-5 years. But I am data-mining, so take the results with a grain of salt. For example, survivorship bias. And maybe the stocks at the top, like Microsoft and Amazon, weren't at the top back then. But, on the other hand, maybe I would have selected them because they were chosen randomly down the list. I think there is clearly some merit to small, equal-weight, total market proxy indices. But whether the outperformance is real, I can't say. I do like the 31-stock SP500 mimic and the last one with the 2^2.4 criterion. The former avoids small-caps, so it mimics its index well. The latter has a good mix of large-, mid-, small-, and micro-caps, so it better resembles the total market, but more modeling will be necessary to check if I wasn't just lucky with my selections. At least the results are quite positive. If they were all failures I'd give up now! There is still hope.
Pet Hog
Executive Member
Executive Member
Posts: 218
Joined: Tue May 28, 2013 4:08 pm

Re: DIY individual stock fund

Post by Pet Hog » Tue Sep 24, 2019 4:45 pm

Given the positive results, I'm going to model one more portfolio, based on the n^2.4 concept. I haven't started the model, but here are the rules.

(1) Because the micro-caps are difficult to model (often with few data because they are new) and because they are volatile and I might not want to hold them for, say 5 years, I'll only consider companies with a market cap of greater than $500 million (selected arbitrarily). According to suredividend there are 2238 of them.

(2) Because I want to avoid data-mining of the FAANGs, and because they highest market cap stocks have less room to grow, I won't consider the most valuable company and I won't consider any other company with a market cap two-thirds as high (another arbitrary choice). This rule eliminates Microsoft, Apple, Amazon, and Google (GOOG and GOOGL). The top remaining stock is Facebook. I consider it number 1 in my list.

(3) The number of remaining stocks is 2238 minus 5, or 2233. I'm going to select 40 stocks (arbitrary) for my portfolio. So I need to find a number n so that 40^n < 2233. Using logs, I get 2.09.

(4) I'll pick the stocks, starting from Facebook as number 1, according to their market caps, in the sequence n^2.09, for n = 1, 2, 3..., 39, 40. I'll take them from the website listed above. I'll then use portfolio visualizer to backtest this portfolio for at least five years (that might necessitate dropping some of the companies from the list).

Those are the rules, now off to work.
Pet Hog
Executive Member
Executive Member
Posts: 218
Joined: Tue May 28, 2013 4:08 pm

Re: DIY individual stock fund

Post by Pet Hog » Tue Sep 24, 2019 6:37 pm

The n^2.09 portfolio, with mega-caps and micro-caps avoided.

The first three columns are n, n^2.09 (ignoring fractions), and n^2.09 + 5 [e.g., Facebook, the first company on this list (n = 1) is actually 6th on the list of richest companies]

01 0001 0006 Facebook (532 billion)
02 0004 0009 Johnson & Johnson (347 billion)
03 0009 0014 Mastercard, Inc. (277 billion)
04 0018 0023 UnitedHealth Group, Inc. (216 billion)
05 0028 0033 Citigroup, Inc. (157 billion)
06 0042 0047 Union Pacific Corp. (116 billion)
07 0058 0063 American Express Co. (98 billion)
08 0077 0082 Chubb Ltd. (71 billion)
09 0098 0103 Boston Scientific Corp. (60 billion)
10 0123 0128 Waste Management, Inc. (48 billion)
11 0150 0155 Activision Blizzard, Inc. (41 billion)
12 0180 0185 Valero Energy Corp. (34 billion)
13 0212 0217 SBA Communications Corp. (28 billion)
14 0248 0253 KKR & Co., Inc. (24 billion)
15 0287 0292 Corning, Inc. (21 billion)
16 0328 0333 Western Digital Corp. (18 billion)
17 0372 0377 Evergy, Inc. (15 billion)
18 0420 0425 Universal Health Services, Inc. (13 billion)
19 0470 0475 Norwegian Cruise Line Holdings Ltd. (11 billion)
20 0523 0528 Packaging Corporation of America (9 billion)
21 0580 0585 Douglas Emmett, Inc. (8 billion)
22 0639 0644 AptarGroup, Inc. (7 billion)
23 0701 0706 East West Bancorp, Inc. (6 billion)
24 0766 0771 Scotts Miracle-Gro Co. (5 billion)
25 0835 0840 Blackstone Mortgage Trust, Inc. (4 billion)
26 0906 0911 Boston Beer Co., Inc. (4 billion)
27 0980 0985 United Therapeutics Corp. (3 billion)
28 1058 1063 Immunomedics, Inc. (3 billion)
29 1138 1143 Inphi Corp. (3 billion)
30 1222 1227 Box, Inc. (3 billion)
31 1309 1314 CareTrust REIT, Inc. (2 billion)
32 1398 1403 Fabrinet (2 billion)
33 1491 1496 Patterson Cos., Inc. (2 billion)
34 1587 1592 Aimmune Therapeutics, Inc. (1.4 billion)
35 1686 1691 Helix Energy Solutions Group, Inc. (1.2 billion)
36 1789 1794 SecureWorks Corp. (1.0 billion)
37 1894 1899 Federal Agricultural Mortgage Corp. (881 million)
38 2003 2008 Emerald Expositions Events, Inc. (732 million)
39 2115 2120 CorePoint Lodging, Inc. (609 million)
40 2230 2235 Entercom Communications Corp. (503 million)

From portfoliovisualizer, monthly returns, dividends reinvested. I'm surprised how well this strategy back-tests, although I'm aware that I wouldn't have picked these stocks back then. Still, quite a random selection of companies. Again, I had to ignore a handful of stocks to allow back-testing for five years.

Image

Image

Image
Pet Hog
Executive Member
Executive Member
Posts: 218
Joined: Tue May 28, 2013 4:08 pm

Re: DIY individual stock fund

Post by Pet Hog » Tue Sep 24, 2019 9:15 pm

I went looking for historical Wilshire 5000 component data, but came up short. I'm sure it's out there somewhere. Found historical SP500 component data on the iShares IVV page. Taking their listing of the components of IVV from August 28 2014, let's see how a simple every-sixteenth-stock selection strategy would have done over the last five years.

016 Intel
032 Home Depot
048 American Express
064 Dow Chemical (merged with DuPont)
080 Anadarko Petroleum (acquired by Occidental Petroleum)
096 Danaher Corp
112 DirecTV (purchased by AT&T for $95/share)
128 Delta Airlines
144 State Street
160 Marathon Oil
176 Noble Energy
192 Chubb
208 Chipotle Mexican Grill
224 Edison International
240 Invesco
256 Carnival Corp
272 Marriott International
288 Whole Foods Market (purchased by Amazon at $42/share)
304 Conagra Brands
320 Cimarex Energy
336 Mattel
352 Under Armour
368 WEC Energy Group
384 F5 Networks
400 Hospira (purchased by Pfizer)
416 International Flavors and Fragrances
432 Scana (purchased by Dominion)
448 Pepco Holdings (merged with Exelon)
464 Pinnacle West
480 Apartment Investment and Management
496 Bemis (merged with Amcor) <--oops, mistakenly used AMC Entertainment in the model

Whenever there was a merger or acquisition I used the data for the remaining company. Like Amazon for Whole Foods. I don't think that's fair, but easier to model. Lot's of energy companies in this list, all poor performers. Bit of a dud portfolio overall. Back to reality!

Image

Image

Image
Post Reply